Immunity

Article

T Cell Allorecognition via Molecular MimicryWhitney A. Macdonald,1 Zhenjun Chen,2 Stephanie Gras,1 Julia K. Archbold,1 Fleur E. Tynan,1 Craig S. Clements,1

Mandvi Bharadwaj,2 Lars Kjer-Nielsen,2 Philippa M. Saunders,2 Matthew C.J. Wilce,1 Fran Crawford,4 Brian Stadinsky,4

David Jackson,2 Andrew G. Brooks,2 Anthony W. Purcell,3 John W. Kappler,4 Scott R. Burrows,5 Jamie Rossjohn,1,6,*and James McCluskey2,6,*1The Protein Crystallography Unit, Department of Biochemistry and Molecular Biology, Monash University, Clayton, Victoria 3800, Australia2Department of Microbiology & Immunology3Department of Biochemistry and Molecular Biology and Bio21 Molecular Science and Biotechnology Institute

University of Melbourne, Parkville, Victoria 3010, Australia4Howard Hughes Medical Institute, Integrated Department of Immunology, National Jewish Health, 1400 Jackson Street, Denver,

CO 80206, USA5Cellular Immunology Laboratory, Queensland Institute of Medical Research and Australian Centre for Vaccine Development,

Brisbane, 4029, Australia6These authors contributed equally to this work

*Correspondence: jamie.rossjohn@med.monash.edu.au (J.R.), jamesm1@unimelb.edu.au (J.M.)

DOI 10.1016/j.immuni.2009.09.025

SUMMARY

T cells often alloreact with foreign human leukocyteantigens (HLA). Here we showed the LC13 T cellreceptor (TCR), selected for recognition on self-HLA-B*0801 bound to a viral peptide, alloreactswith B44 allotypes (HLA-B*4402 and HLA-B*4405)bound to two different allopeptides. Despite exten-sive polymorphism between HLA-B*0801, HLA-B*4402, and HLA-B*4405 and the disparate se-quences of the viral and allopeptides, the LC13 TCRengaged these peptide-HLA (pHLA) complexes iden-tically, accommodating mimicry of the viral peptideby the allopeptide. The viral and allopeptides adop-ted similar conformations only after TCR ligation,revealing an induced-fit mechanism of molecularmimicry. The LC13 T cells did not alloreact againstHLA-B*4403, and the single residue polymorphismbetween HLA-B*4402 and HLA-B*4403 affected theplasticity of the allopeptide, revealing that molecularmimicry was associated with TCR specificity.Accordingly, molecular mimicry that is HLA andpeptide dependent is a mechanism for human T cellalloreactivity between disparate cognate and alloge-neic pHLA complexes.

INTRODUCTION

Clonally distributed ab T cell receptors (TCR) corecognize

specific antigenic peptides bound to polymorphic human leuko-

cyte antigens (HLA) of the major histocompatibility complex

(MHC) (Davis et al., 1998; Rudolph et al., 2006). HLA polymor-

phism ensures that the HLA molecules from different haplotypes

can bind a broad sample of self and microbial peptide antigens

necessary to mediate adaptive immunity (Parham and Ohta,

1996). Developing T cells in the thymus are selected for weak

recognition of one or more of the many self-peptide-HLA

I

complexes (Bevan and Hunig, 1981; Hogquist et al., 1993)

generating a large repertoire of T cells, each expressing indi-

vidual TCRs (Fink and Bevan, 1995). Inherent structural plasticity

of the TCR contributes to chance improvements in recognition of

novel peptide-HLA complexes (pHLA) that are generated when

self-peptides are replaced with foreign peptides during infection

(Garcia et al., 1998, 1999; Rudolph et al., 2006). This recognition

triggers effector immunity by responsive T cells.

Despite pHLA diversity and TCR plasticity, ab-T cell

responses remain exquisitely specific (Archbold et al., 2009)

and are developmentally restricted to recognizing host (self)

HLA (Jameson et al., 1995; Zinkernagel and Doherty, 1974),

with the exception of minor subpopulations like NKT cells

(Borg et al., 2007). This ‘‘genetic restriction’’ of MHC-directed

T cell immunity means that T cells recognize only cognate

antigen presented by one of the host HLA molecules in which

they developed (also termed MHC restriction) (Zinkernagel and

Doherty, 1974). This ‘‘law’’ of immunology is a defining paradigm

of antigen-specific T cell immunity (Garboczi and Biddison,

1999).

Surprisingly, some T cells break the ‘‘law’’ of MHC restriction

(Sherman and Chattopadhyay, 1993) by directly reacting with

‘‘foreign’’ HLA molecules from unrelated (allogeneic) individuals.

HLA polymorphism involving just one amino acid, or up to 30 or

more residues, can induce an immune response toward trans-

planted cells, the severity of which is variable. Thus, some HLA

mismatches lead to worse transplant outcomes than others,

so-called taboo mismatches (Doxiadis et al., 1996; Kawase

et al., 2007). For instance, mismatching across closely related

HLA allotypes such as HLA-B*4402 and HLA-B*4403 provokes

vigorous T cell alloreactivity (Mifsud et al., 2008) associated

with transplant rejection (Fleischhauer et al., 1990) and acute

graft-versus-host disease (Keever et al., 1994) after haemo-

poietic stem cell transplantation, despite the broadly similar

peptide repertoires of these allotypes (Macdonald et al., 2003).

In contrast, highly divergent HLA mismatches may paradoxically

have a better outcome in some transplant settings (Heemskerk

et al., 2007). Regardless, T cell alloreactivity is responsible for

much of the morbidity and mortality associated with tissue trans-

plantation, including graft-versus-host disease (Afzali et al.,

mmunity 31, 897–908, December 18, 2009 ª2009 Elsevier Inc. 897

Immunity

T Cell Alloreactivity Mediated by Molecular Mimicry

2007), making this unexplained contradiction to the phenom-

enon of MHC restriction of great clinical importance.

The paradox of alloreactivity has remained a mystery for more

than three decades (Archbold et al., 2008b; Droge, 1979; Lechler

and Lombardi, 1990), including the reasons for the high

frequency of these T cells (Lindahl and Wilson, 1977) and

whether the peptide or the HLA molecule is more important in

driving T cell alloreactivity (Bevan, 1984; Matzinger and Bevan,

1977). The HLA-centric model of alloreactivity considers that

T cells concentrate on the polymorphic HLA residues irrespec-

tive of the bound peptide. For instance, the alloreactive murine

2C TCR adopts two very different binding orientations when

bound to its host selecting-pMHC ligand versus an allogeneic

pMHC target ligand, focusing instead on a mixture of allogeneic

MHC differences and new peptide contacts (Colf et al., 2007;

Rossjohn and McCluskey, 2007). In contrast, the peptide-centric

theory of allorecognition implies that the TCR exploits the simi-

larities between the allogeneic and self-HLA molecule

(‘‘mimicry’’) and recognizes the new set of endogenous peptides

as foreign. Additionally, molecular mimicry is considered to

underpin numerous T cell autoimmune disorders but has never-

theless been difficult to establish given the explicit requirement

of the TCR to corecognize the antigens as well as the HLA mole-

cules. Moreover, limited evidence so far suggests that T cell al-

lorecognition is peptide centric (Archbold et al., 2008a; Colf

et al., 2007; Reiser et al., 2000; Speir et al., 1998) and implicates

polyspecificity as a mechanism leading to the high frequency of

alloreactive T cells (Felix et al., 2007). However, it is still unclear

whether dual recognition of disparate cognate and allogeneic

pHLA by a single TCR can involve similar binding modes, namely

operating via molecular mimicry (Archbold et al., 2008b; Ross-

john and McCluskey, 2007). Here we show that molecular

mimicry can underpin human T cell alloreactivity.

RESULTS

Peptide-Dependent Alloreactivity of LC13 T CellsTo investigate the molecular basis of natural human T cell allor-

eactivity, we examined the prototypic TCR termed LC13 that

recognizes the immunodominant HLA-B*0801-restricted

epitope, FLRGRAYGL from EBNA 3A of Epstein-Barr virus

(EBV) (Argaet et al., 1994; Burrows et al., 1994). LC13 also allor-

eacts with HLA-B*4402 and HLA-B*4405, related allotypes that

differ from each other by only one residue but differ from HLA-

B*0801 by 24 and 25 amino acids, respectively.

Alloreactivity can be either dependent or independent of the

HLA-bound peptide (Heath et al., 1989, 1991; Smith et al.,

1997a, 1997b). Therefore, we examined whether LC13 allore-

cognition of HLA-B*4405 required a specific peptide(s). Presen-

tation of the HLA-B*4405 alloantigen was examined in transfec-

tants of the class-I-HLA-deficient mutant lymphoblastoid cell line

(LCL) C1R and the TAP-deficient T2 cell line (Alexander et al.,

1989). The C1R.B*4405 cells, but not the parental C1R cells,

were lysed by LC13 cytotoxic T-lymphocyte (CTL), indicating

constitutive presentation of an allogeneic ligand by these cells

(Figure 1A). However, coexpression of the viral TAP inhibitor

ICP47 essentially abolished allorecognition of C1R-B*4405 by

LC13 (Figure 1A), indicating TAP dependence of this allogeneic

898 Immunity 31, 897–908, December 18, 2009 ª2009 Elsevier Inc.

ligand. Exogenous loading of C1R-B*4405-ICP47 cells with viral

peptide restored recognition by an antiviral CTL clone (DM1)

(Archbold et al., 2009) but did not restore killing by LC13 CTL

(Figures 1A and 1B). The T2.B*4405 cell line was not recognized

by the human T cell line Jurkat coexpressing the LC13 ab TCR

and human CD8ab genes (LC13.Jurkat) (Beddoe et al., 2009).

Stabilization of ‘‘empty’’ HLA-B*4405 molecules with a HLA-

B*4405-binding peptide (DPa-peptide) did not sensitize the

T2.B*4405 cells for recognition by LC13.Jurkat (Figure 1C).

Notably, the T2.B*0801 and C1R.B*0801 cell lines loaded with

exogenous FLRGRAYGL viral peptide (‘‘virotope’’) activated

LC13.Jurkat (Figures 1C and 1D), as did C1R.B*4402 and

C1R.B*4405 transfectants (Figure 1D). Collectively, these data

indicate that the alloreactivity of the LC13 TCR behaved in a

peptide-dependent manner.

Identification of a Candidate Allopeptide Presentedby HLA-B*4405A major hurdle in understanding the basis of alloreactivity is the

identification of authentic antigenic peptides (the allopeptide[s])

bound to the allogeneic HLA molecule. Murine examples of allor-

eactive T cells have been the most informative to date, including

the alloreactive BM3.3 TCR (Reiser et al., 2000) and the 2C TCR

(Colf et al., 2007; Speir et al., 1998) for which pMHC allopeptide

structures are solved. However, pathogen-derived cognate

ligands for the BM3.3 and 2C T cells remain unknown.

To identify a candidate LC13 allopeptide(s), we generated

insect cells expressing individual baculoviral constructs from

a library of HLA-B*4405 molecules covalently complexed with

randomized peptides. Infected insect cells were screened for

interaction with recombinant, bivalent LC13 TCR (Crawford

et al., 2006). Repeated rounds of sorting allowed expansion of

HLA-B*4405-positive cells expressing a ligand that bound

LC13 TCR (Figure 2A). Peptide insert sequences were obtained

from 36 positive clones with 30 of these encoding the peptide

EEYLKAWTF. Searching the human proteome for analogs of the

EEYLKAWTF ‘‘mimotope’’ peptide identified two high-scoring

matches (expect values of 283 and 65, respectively), each of 9

residues (EESLKDWYF and EEYLQAFTY) and therefore repre-

senting a potential natural ‘‘allotope.’’ These peptides shared

66% (6/9 identical residues) with the mimotope and possessed

the P2E, P9Y/F anchor residues, features of B44-binding

peptides. The peptide EESLKDWYF is derived from an ATPase

but little is known about its physiologic role and expression.

The peptide EEYLQAFTY is derived from an ATP binding

cassette protein ABCD3 involved in transport of fatty acids into

the peroxisome.

The ABCD3 Allotope Is an Authentic AlloantigenRecognized by LC13We next examined recognition of the EESLKDWYF or EEYL-

QAFTY peptides by LC13.Jurkat cells (CD8+) and LC13 CTL.

The EESLKDWYF peptide did not activate LC13.Jurkat cells

and was not examined further because we conclude that this

is not a bona fide alloligand for LC13 (not shown). In contrast,

both the mimotope and EEYLQAFTY (hereafter allotope)

peptides specifically sensitized exogenously loaded T2-B*4405

cells (Figure 2B, middle) and C1R-B*4405 cells expressing

ICP47 (Figure 2B, right) for lysis by LC13 CTL.

A

C

B

D

*

**

**

0 40 80 0 50 100

Figure 1. Allorecognition of HLA-B*4405 Is Peptide Dependent(A and B) The LC13 CTL clone (A) and the HLA-B*4405-restricted CTL clone DM1 (B), specific for the EBV peptide EENLLDFVRF, were tested for killing of the

class-I-HLA-negative mutant cell line C1R and its stably transfected derivatives expressing HLA-B*4405 in the presence and absence of the TAP-inhibitor ICP47.

Cytotoxicity at four effector:target ratios is shown.

(C) Activation of LC13.Jurkat cells by the TAP-deficient, T2 cell line expressing low levels of HLA-B*4405 (T2.B*4405) and T2 cells expressing HLA-B*0801

(T2.B*0801). T cell activation is shown on the y axis as the percentage of CD69-positive cells among the GFP-positive cells.

(D) Activation of LC13.Jurkat cells by the TAP-competent C1R.B*4402 (B*4402), C1R.B*4405 (B*4405), and C1R.B*0801 cell lines as above. C1R.B*0801 trans-

fectants are transformed with a strain of EBV containing a mutation in FLRGRAYGL.

Immunity

T Cell Alloreactivity Mediated by Molecular Mimicry

To determine whether the allotope is naturally presented, the

impact of super transfection and knockdown of the ABCD3

gene was studied in cells naturally presenting B*4405-restricted

alloantigen to LC13. Super transfection of the ABCD3 gene into

C1R.B*4405 cells resulted in a modest increase in constitutive

activation of LC13.Jurkat by C1R.B*4405 and C1R.B*4402

cells (Figure S1 available online). A specific RNAi construct

was also used to knock down the natural, endogenous expres-

sion of ABCD3 in Ag-presenting cells (Figure 2C). Real-time

PCR assays of RNA expression showed that the ABCD3 allo-

tope RNAi reduced mRNA expression by >80% and con-

firmed the specificity of the RNAi constructs (semiquantitative

RT-PCR inset, Figure 2C and Figure S2). Mock treatment of

cells with irrelevant mb-actin RNAi had no impact on LC13

allorecognition (Figure 2C, middle panel histograms). In contrast,

introduction of the ABCD3 RNAi into the C1R.B*4405 cells

specifically reduced constitutive activation of LC13.Jurkat

T cells by nearly 50% (p < 0.01) (Figure 2C, right panel

histograms). Addition of exogenous allotope to the knocked

I

down Ag-presenting cells restored full activation of the

LC13.Jurkat T cells (Figure 2C). These data indicate that the

ABCD3 allotope is an authentic, natural alloantigen recognized

by the LC13 TCR.

Molecular Mimicry Underpins LC13 AlloreactivityTo understand the structural basis of the LC13 TCR alloreactiv-

ity, we determined the structures of the LC13 TCR in complex

with the HLA-B*4405 allotope and mimotope complexes to

2.6 A and 2.7 A resolution, respectively (Table 1, Tables S1

and S2). These structures were compared with the LC13-viro-

tope complex (Kjer-Nielsen et al., 2003).

The structure of the LC13 TCR-allotope complex was very

similar to the mimotope complex with a root mean square devi-

ation (rmsd) of 0.27 A over the entire complex, and remarkably,

both complexes (Figures 3A and 3B) were very similar to that of

the LC13 TCR-virotope complex (Figure 3C; Kjer-Nielsen et al.,

2003) (rmsd between the allotope and mimotope complex versus

the virotope complex was 0.87 A and 0.77 A, respectively). This

mmunity 31, 897–908, December 18, 2009 ª2009 Elsevier Inc. 899

A

B

C

Figure 2. Identification of an Authentic Allo-

ligand Recognized by LC13

(A) Baculovirus-infected insect SF9 cells, each ex-

pressing a unique peptide-B*4405 complex on

their surface, were costained with anti-b2Micro-

globulin mAb (y axis) and multimeric recombinant

LC13 TCR (x axis). LC13 staining-SF9 cells were

iteratively expanded and sorted. Sequencing of

viral constructs consistently identified a single

mimotope ligand, EEYLKAWTF.

(B) LC13 recognizes exogenous mimotope and al-

lotope peptides presented by TAP-deficient cells.

Dose-dependent recognition of exogenous allo-

tope (triangles) or mimotope (circles) peptide.

Cytotoxicity is reported as percent specific lysis

by LC13 CTL.

(C) Knockdown of C1R.B4405 recognition by LC13

after RNA interference of ABCD3. The homozy-

gous HLA-B*0801 LCL, Bm.wil, and C1R.B*4405

cell lines were treated with (+) and without (�)

ABCD3 or mouse b-actin RNAi constructs and

constitutive presentation of alloantigen was as-

sayed in the absence (open bars) or presence of

the virotope peptide (Viro) or allotope peptide

(Allo). (NS, nonsignificant; **p < 0.01). Inset shows

an electrophoresis gel photo of PCR amplification

of ABCD3 cDNA from C1R.B4405 cells in the pres-

ence (KO) and absence (WT) of specific ABCD3

RNAi treatment.

Immunity

T Cell Alloreactivity Mediated by Molecular Mimicry

similarity is reflected in the close superposition of the LC13 TCR

in these complexes and their identical 60� docking modes across

the long axis of the HLA (Figure 3D). Accordingly, the LC13 TCR

location over the C terminus of the HLA-B*4405 antigen-binding

cleft mimicked the C-terminal docking of the LC13 TCR on

the HLA-B*0801-virotope complex (Kjer-Nielsen et al., 2002a,

2003). The total buried surface area (BSA) at the allotope, mimo-

tope, and virotope complexes were all z2300 A2 and moreover,

the shape complementarity at the virotope, allotope, and mimo-

tope interfaces with LC13 was very similar (0.59, 0.64, and 0.60,

respectively).

Both the Va and Vb domains of the LC13 TCR contributed

roughly equally to the interfaces of the allotope, mimotope,

and virotope complexes (range: Va, 51.4%–56.2%, Vb,

900 Immunity 31, 897–908, December 18, 2009 ª2009 Elsevier Inc.

43.8%–48.6%), indicating that the LC13

alloreactivity is not driven by a skewed

usage of the V domains at the TCR-

pMHC interface unlike other alloreactive

complexes (Colf et al., 2007; Reiser

et al., 2000). Indeed, the number and

nature of the LC13 TCR interactions with

the pHLA B*4405 in the allotope and

mimotope complexes were also very

similar to those of the LC13 TCR-virotope

complex (allotope-mimotope-virotope:

146-160-135 van der Waals [v.d.w.] inter-

actions, 15-13-14 H bonds, and 1 salt

bridge each; Table S2). Accordingly, the

LC13 TCR adopted a strikingly similar

footprint on the allogeneic HLA-B*4405-

allotope, HLA-B*4405-mimotope, and cognate HLA-B*0801-

virotope complex.

Mimicry of the TCR Footprints and Specific InteractionsAlthough the overall docking modes between the LC13 TCR-

allotope, mimotope, and virotope complexes were very similar,

this does not confirm molecular mimicry at the molecular level.

Therefore, we analyzed the individual contacts of the LC13

TCR with each of these three complexes. The relative contact

footprints of the complementarity determining region (CDR)

loops at the LC13 TCR-pHLA interfaces were also very similar

(Figure 3, bottom). Hence, to varying extents, all the CDR loops

of the LC13 TCR contributed to virotope, allotope, and mimo-

tope interactions, with only modest differences between them

Table 1. Data Collection and Refinement Statistics

Data Collection

Statistics

LC13-HLA

B4405allo

LC13-HLA

B4405mimo

Temperature 100K 100K

Space group C2 C2

Cell dimensions

(a,b,c) (A, �)

142.52, 54.24,

121.77; b = 114.43

223.12, 53.22,

143.20; b = 102.39

Resolution (A) 50-2.70 (2.80-2.70)a 50-2.60 (2.69-2.60)

Total number of

observations

70,144 159,863

Number of unique

observations

21,530 (1,278) 50,197 (4,976)

Multiplicity 3.2 (2.0) 3.2 (3.0)

Data completeness (%) 92.0 (55.6) 97.8 (97.8)

I/sI 20.1 (2.4) 12.5 (2.3)

Rmergeb (%) 5.9 (25.6) 8.5 (41.1)

Refinement Statistics

Nonhydrogen atoms

Protein 6,657 13,316

Water 27 124

Resolution (A) 2.70 2.60

Rfactorc (%) 19.7 22.1

Rfreec (%) 26.9 27.8

Rms deviations from

ideality

Bond lengths (A) 0.009 0.006

Bond angles (�) 1.202 0.926

Ramachandran plot (%)

Most favored region 87.1 90.5

Allowed region 12.0 9.0

Generously allowed region 0.6 0.4a Values in parentheses are for highest-resolution shell.b Rmerge = S j Ihkl - < Ihkl > j / SIhkl.c Rfactor = Shkl j j Fo j - j Fc j j / Shkl j Fo j for all data except z5% that were

used for Rfree calculation.

Immunity

T Cell Alloreactivity Mediated by Molecular Mimicry

(the rmsd of the respective CDR2a, CDR3a, CDR1-3b loops

within the complexes was <0.45 A) (Table S2). One slight differ-

ence was the positioning of the CDR1a loop between the allo-

tope and virotope complexes (rmsd approximately 1.0 A) (not

shown). Overall, the conformational changes of the LC13 TCR

in forming interactions with the virotope complex (Kjer-Nielsen

et al., 2002b, 2003) are mirrored in the interactions with the

HLA-B*4405-allotope and mimotope structures, despite the

differences in the antigenic peptide sequences.

The CDR1a makes conserved contacts via Gly29a, Thr30a,

and Tyr31a with the a2 helix of HLA-B*4405. The Arg62 of

HLA-B*4405 contacts the P1 residue of the allotope, and

Tyr159 of HLA-B*4405 contacts Thr30a (Figure 4A). Thr30a

and Tyr31a enveloped the ‘‘gatekeeper’’ residue Gln155, which

changed conformation upon LC13 TCR ligation in all three

complexes. Accordingly, the CDR1a loop played a similar role

in the overall contribution to interactions in the virotope complex

(18.2%) when compared to the allotope and mimotope

complexes (16%).

I

The CDR2a loop of the LC13 TCR contributed equally to the

interface in the allotope, mimotope, and virotope interactions

(approximately 8%–9%; Figures 3A–3C) and interacted via

His48a, Leu50a, Ser52a, and Val55a with the a2 helix of HLA-

B*0801 and HLA-B*4405, nestling against the long side chains

of Arg151 and Glu154 (Figure 4B). These conserved interactions

are mediated predominantly via vdw interactions (Figure 4B;

Table S2).

The CDR1b loop minimally participated in the pHLA interac-

tions (Figure 3: Table S2). In contrast, the CDR2b loop contrib-

uted equally to the interface in the allotope, virotope, and mimo-

tope interactions (approximately 13%–14%, Figure 4C), through

conserved contacts via Tyr48b, Gln50b, Asn51b, Glu52b, and

Leu55b and the a1 helix of HLA-B*0801 (residues 72–79) and

HLA-B*4405 (residues 72–83). This network of polar-mediated

contacts includes one conserved salt bridge between Glu52b

and Arg79 (Figure 4C). Ala53b makes an additional contact

with Arg75 of HLA-B*4405. Interestingly, the CDR2b loop inter-

acted with HLA-B*4405 position 83, a polymorphic site between

HLA-B*4405 (Arg83) and HLA-B*0801 (Gly83) (Figure 4C).

However, previous mutagenesis has indicated that the CDR2b

loop plays a minor energetic role in the LC13 TCR-HLA-B8-viro-

tope interaction and is therefore unlikely to be important in allo-

geneic recognition (Borg et al., 2005).

The CDR3a and CDR3b regions contributed approximately

equally at the allotope, virotope, and mimotope interfaces

(18.9%–21.4% and 24%–25.6%, respectively; Figure 3). The

CDR3b loop dominated contacts with the respective peptides

(discussed below), whereas the CDR3a loop played a larger

role in interacting with the HLA heavy chain a1 helix and also

forming interactions with Leu94a and Gln155 of the a2 helix

that are conserved across all three complexes (Figure 4D). The

conserved interactions between the three complexes also

included Gly96a to the aliphatic base of Arg62; Gly97a to Ile66;

and contacts via Thr98a and Tyr100a (Figure 4D). Ser99a makes

a new B*4405 contact not present in HLA B*0801. The CDR3b

loop, which abutted the CDR3a loop and sits centrally above

the Ag-binding cleft, mediated contacts with the a1 and the a2

helix of the HLA, in which Gln98b and Tyr100b protruded into

the cleft to form conserved interactions along with Leu96b and

Gly97b.

Accordingly, a very high degree of mimicry of the cognate

HLA-B*0801-virotope underpinned LC13 TCR interactions

conserved across the HLA-B*4405-allotope and mimotope

complexes.

Peptide-Dependent Molecular MimicryGiven the differences in the sequences between the cognate vi-

rotope, allotope, and mimotope, it was unclear, a priori, whether

the peptide-mediated interactions made by the LC13 TCR would

be similar between all three complexes. Therefore, we compared

the mode of binding of the LC13 TCR to the different peptides.

Upon superposition, the rmsd of the HLA-bound cognate

peptide with respect to the bound allotope and mimotope was

0.79 A and, thus, within the ternary complexes, the peptides

adopted similar conformations within the respective Ag-binding

cleft (Figure 5). Although the LC13 TCR also interacted with

the N-terminal region of the allotope and mimotope peptides

(Figures 5A and 5B), the extensive interactions with the

mmunity 31, 897–908, December 18, 2009 ª2009 Elsevier Inc. 901

A B C D

Figure 3. Footprint of LC13 TCR in Complex with HLA-B*4405-Allotope, HLA-B*4405-Mimotope, and HLA-B*0801-Virotope

(A–C) Ribbon representation of the LC13 TCR in complex with HLA-B*4405-allotope (A), HLA-B*4405-mimotope (B), and HLA-B*0801-virotope (C). TCR a chain is

in pale pink; the b-chain is in pale blue, HLA-B*4405 is in dark gray, HLA-B*0801 in pale gray, the peptide is in stick format, colored marine for the allotope (A),

orange for the mimotope (B), and purple for the virotope (C). Residues contacted by the CDR loops are colored in red (CDR1a), green (CDR2a), blue (CDR3a),

orange (CDR1b), pink (CDR2b) and cyan (CDR3b) in the three complexes.

(D) Superposition of LC13 TCR (allotope, blue-green; virotope, purple) in complex with HLA-B*4405-allotope and HLA-B*0801-virotope. The surface represen-

tation of HLA-B*0801-virotope at the bottom of the panel shows the CDR loops of the LC13 TCR in complex with HLA-B*4405-allotope (in marine) and with HLA-

B*0801-virotope (in purple). The black spheres represent the orientation on the Va and Vb chains of the LC13 TCR calculated by center of mass.

Immunity

T Cell Alloreactivity Mediated by Molecular Mimicry

C-terminal (P6–8) region of the peptides, known to be critical for

recognition of the virotope (Kjer-Nielsen et al., 2003), were highly

similar in all three complexes. Namely, the C-terminal residues of

both the allotope (P6-Ala, P7-Phe, P8-Thr) and mimotope

(P6-Ala, P7-Trp, P8-Thr) include a bulky aromatic side chain at

P7 that was flanked by small amino acids, a feature important

for LC13 recognition of the virotope (P6-Ala, P7-Tyr, P8-Gly)

(Figures 5A–5C; Kjer-Nielsen et al., 2003). Consequently,

mimicry in this region underpins how the LC13 TCR interacted

with the P6–P8 region of the allotope and the mimotope

peptides. The small P6 and P8 residues enabled the P7-aromatic

to protrude within a central pocket of the LC13 TCR, as well as

contributing to specificity-governing interactions with the LC13

TCR. Namely, the P6-Ala made a conserved interaction with

Leu94a of CDR3a and Ala99b of CDR3b, and the backbone of

P6-Ala formed a conserved H bond with Gln98b of the CDR3b

loop (Figures 5A–5C). Despite the different P8 side chains

902 Immunity 31, 897–908, December 18, 2009 ª2009 Elsevier Inc.

between the virotope (P8-Gly) and allotope/mimotope (P8-Thr),

Tyr100b of the CDR3b loop formed a conserved H bond with

the backbone of P8 (Figures 5A–5C). The aromatic structures

of the P7 residues were each sandwiched between Tyr31a and

Tyr100b and contacted Ala99b (Figures 5A–5C). The P7-TyrOH

of the virotope formed critical water-mediated interactions with

His33a and His48a (Figure 5C); however, because of the differ-

ences at this position in the mimotope (P7-Trp) and allotope

(P7-Phe), these water-mediated interactions were absent in

these complexes (Figures 5A and 5B). The P7-Trp of the mimo-

tope formed a H-bond with Tyr31a (Figure 5B). Interestingly,

mutating P7-Phe to P7-Tyr of the allotope increased recognition

by the LC13 TCR to levels comparable to that of the cognate

interaction (data not shown).

These findings also underscore the lack of recognition of the

EESLKDWYF candidate peptide identified in the BLASTp

search, because the bulky side chain of P8-Tyr in this ligand

A B

C D

Figure 4. Conserved LC13 TCR Contacts with Cognate and Allogeneic Ligands

Contacts made by the LC13 TCR with the HLA-B*4405-allotope complex. The CDR loops of the LC13 TCR are shown in stick format; the allotope is blue-green

and the HLA-B*4405 is dark gray. Interactions between LC13 TCR and HLA-B*4405 that are conserved between LC13 TCR and the HLA-B*0801-virotope are

colored in red, and those specific to HLA-B*4405 complex are colored blue. Shown are (A) CDR1a contacts, (B) CDR2a contacts, (C) CDR2b contacts, and (D)

CDR3a and CDR3b contacts. Gln155 changes conformation upon ligation and is colored cyan in the nonligated state.

Immunity

T Cell Alloreactivity Mediated by Molecular Mimicry

would sterically obstruct recognition of the P7 aromatic crucial to

recognition of the B*4405-virotope, allotope, and mimotope

complexes. Accordingly, in addition to the mimicry between

the surface topology of the HLA-B*0801 and B*4405 heavy

chains, substantial mimicry of the HLA-B*0801-restricted viro-

tope underscored how the LC13 TCR interacted with the critical

C-terminal region of the HLA-B*4405-restricted allotope and

mimotope.

Alloreactivity Discriminates between RelatedB44 AllotypesLC13 alloreacts with HLA-B*4402 and HLA-B*4405, but surpris-

ingly not with HLA-B*4403 (Burrows et al., 1994, 1995, 1997).

Therefore, we tested LC13 recognition of phytohaemagglutinin

(PHA) blast cells expressing either HLA-B*4405, HLA-B*4402,

or HLA-B*4403 after adding exogenous mimotope or allotope

peptide (Figure 6A). Consistent with the defined specificity of

LC13 (Burrows et al., 1997), the HLA-B*4403+ cells were not

recognized at physiological concentrations of the mimotope

I

peptide. This might partly reflect lower binding of the allotope

peptide to B*4403 (not shown). Interestingly, the HLA-B*4405+

cells presented both peptides more efficiently than did HLA-

B*4402. Notably, the allotope and mimotope peptides com-

plexed with HLA-B*4405 were recognized at even lower peptide

concentrations than the cognate FLRGRAYGL virotope peptide,

presented by HLA-B*0801+ PHA blasts (Figure 6A). This differ-

ence appeared to result from differential T cell recognition of

these ligands rather than differences in peptide-HLA binding

affinity, as shown by the fact that cross-blocking of pHLA-

tetramer staining of LC13-like T cells confirmed the binding hier-

archy HLA-B*4405-mimotope > HLA-B*4405-allotope > HLA-

B*0801-virotope tetramer (Figure S3).

We then tested whether fine specificity of alloreactivity and

pHLA-tetramer staining correlated with the affinity of the LC13

TCR-pHLA interaction via surface plasmon resonance (SPR)

studies. The LC13 TCR bound to the HLA-B*4405-mimotope

complex with comparable affinity to the HLA-B*4402-mimotope

complex (Kd = 1.5 mM and 1.3 mM, respectively) but interacted

mmunity 31, 897–908, December 18, 2009 ª2009 Elsevier Inc. 903

A B C

Figure 5. Mimicry in Peptide-TCR Contacts

Contacts between the LC13 TCR and the allotope EEYLQAFTY (dark blue) (A), the mimotope EEYLKAWTF (orange) (B), and the virotope FLRGRAYGL (purple) (C).

The peptide is represented in stick format and the LC13 TCR side chains involved in peptide contact are shown. Colors: CDR1a, red; CDR3a, blue; and CDR3b,

cyan. The LC13 TCR makes conserved contacts with the allotope (A), mimotope (B), and virotope (C) at positions P6–P8. In addition, the LC13 TCR makes some

water-mediated contacts (red dash lines) via His33a and His48a with the Tyr7 of the virotope (C). The interactions made by the LC13 TCR with P4-Leu of both the

allotope and mimotope peptides were exclusively via residues from the CDR3a loop and collectively this resulted in a greater contribution of the CDR3a loop in

contacting the mimotope (48.5%) and allotope (47%) when compared to the CDR3a-mediated contacts of the virotope (37%). The CDR1a loop of the LC13 TCR

contacts P3 of the mimotope.

Immunity

T Cell Alloreactivity Mediated by Molecular Mimicry

only very weakly with the HLA-B*4403-mimotope complex (Kd >

200 mM) (Figure 6B; Table S3 and Figure S4). The affinity of the

LC13 TCR for the cognate virotope complex fell between these

values (Kd �10–15 mM) (Kjer-Nielsen et al., 2003).

The LC13 TCR bound the HLA-B*4405-allotope complex with

a higher affinity than the HLA-B*4402-allotope complex (Kd

�49 mM versus�189 mM) (Figure 6C; Table S3), whereas binding

of the LC13 TCR to the HLA-B*4403-allotope complex was very

weak (Kd > 200 mM) (Figure 6C; Figure S4 and Table S3), consis-

tent with the lack of LC13 T cell alloreactivity on HLA-B*4403

(Figure 6A; Burrows et al., 1994, 1995, 1997). Taken together,

the cellular recognition and SPR binding studies reflect the

intrinsic ability of LC13 TCR to discriminate closely related

HLA-B44 allotypes presenting either the mimotope or allotope

determinants, despite the hidden nature of the polymorphic

HLA residues that distinguish these allotypes.

Molecular Basis for Fine Specificity of AlloreactivityTo better understand how the LC13 TCR could discriminate

between HLA-B*4405, HLA-B*4403, and HLA-B*4402 when

bound to the allotope and its mimotope, we determined the

structures of the six binary pHLA-B44 complexes to high reso-

lution (Table S1 and Figure S5). These three allotypes differ

from each other by only 1–2 amino acids in nonexposed posi-

tions unable to directly impact on TCR recognition (residue 116

located in the F pocket and/or residue 156 on the a2 helix, D/E

pocket). Superposition of the HLA-B*4405, HLA-B*4402, and

HLA-B*4403-mimotope complexes revealed virtually no move-

ment of the Ag-binding clefts and modest movements in the

peptide attributable to the interactions between the polymor-

phic residues and the bound peptide (Figure S5). Similarly,

superposing the HLA-B*4405, HLA-B*4402, and HLA-B*4403-

allotope complexes revealed virtually no movement of the

MHC-I heavy chain or the peptide (rmsd 0.18 A and 0.15 A,

respectively).

904 Immunity 31, 897–908, December 18, 2009 ª2009 Elsevier Inc.

The conformation of the allotope and mimotope in their

respective LC13-ternary complexes were very similar (rmsd z0.31 A) (Figure 6D). However, when the structures of the HLA-

B*4405-allotope and HLA-B*4405-mimotope were compared

in the absence of LC13 TCR ligation, the conformation of the

allotope and the mimotope differed markedly (rmsd 0.94 A)

whereas the HLA Ag-binding cleft adopted the same conforma-

tion (rmsd 0.20 A) (Figure 6E). Namely, there were major differ-

ences in the conformation of the peptides between P3-Tyr and

P7-Trp/Phe, where, for example, P5-Gln of the allotope and

the P5-Lys of the mimotope pointed down or upwards from

the Ag-binding cleft, respectively.

This observation revealed that conformational plasticity of the

allotope and mimotope play an important role in the alloresponse

(compare Figures 6D and 6E). For example, upon ligation, the

P3-Tyr of the allotope rotated downwards to avoid steric clashes

with Gln155, but nevertheless maintained an H bond with

Asp156 and formed an additional H bond to Asp114 (Figure 6F).

Additionally, the mimotope is significantly remodeled upon LC13

TCR ligation (Figure 6G). Namely, the CDR1a and CDR3a loops

pushed down the central region of the mimotope (Figure 5B),

causing P4-Leu to be shifted aside and P5-Lys to flip downwards

into the Ag-binding cleft, forming an H bond to Tyr116 and salt

bridging to Asp114 and Asp156 (Figure 6G). The movement of

Gln155 (Figure 4D) also caused a remodeling of P3-Tyr and

P7-Trp, where the P3-Tyr rotated downwards to form a H bond

with Asp156 and the P7-Trp side chain flips 180� to establish

more contacts with the LC13 TCR. Collectively, the mimotope

and allotope mimicked the conformation of the virotope only in

the ligated state and thus peptide-dependent molecular mimicry

is ‘‘forced’’ by the LC13 TCR (Figures 6D and 6E). However, the

LC13 TCR-induced plasticity of the mimotope and allotope

would be disfavored in HLA-B*4403 as a result of Leu156.

Namely, similar plasticity of the mimotope would result in a

buried and uncompensated charge at P5-Lys. Regarding the

A

B C

D E F G

Figure 6. Fine Specificity of the Alloreaction

(A) LC13 recognizes the mimotope and allotope peptides presented by HLA-B*4402 and HLA-B*4405 but not HLA-B*4403. LC13 CTL cytotoxicity of PHA-stim-

ulated T cell lines expressing either HLA-B*4405, HLA-B*4402, or HLA-B*4403. PHA blasts downregulate MHC-I and lose their capacity to be lysed by the LC13

CTLs allowing them to be used as exogenous peptide-presenting targets. Dose response of cytotoxicity on allotope (triangles) or mimotope (circles) peptide.

Lysis of HLA-B*0801-positive PHA blasts loaded with the virotope (squares) is also shown.

(B and C) LC13 TCR binding of HLA-B*4405, HLA-B*4402, and HLA-B*4403 in complex with the mimotope (B) and the allotope (C) as determined by SPR.

(D) Superposition of the allotope (blue-green) and the mimotope (orange) bound to the HLA-B*4405 in the LC13 TCR ligated state.

(E) Structure of the allotope (pink) superposed on the mimotope (green) in complex with HLA-B*4405 but unliganded by LC13.

(F) Conformational change of the allotope in the nonligated (pink) and LC13 TCR-ligated state (marine).

(G) Superposition of the mimotope in complex with HLA-B*4405 both unliganded (green) and liganded (orange) to the LC13 TCR. During the LC13 TCR ligation,

the mimotope undergoes a structural change with the flipping of Lys5. The polymorphic HLA positions (Tyr116 and Asp156 in HLA-B*4405) and the conserved

Asp114 are shown in stick format.

Immunity

T Cell Alloreactivity Mediated by Molecular Mimicry

HLA-B*4403-allotope complex, movement of P3-Tyr would

result in its hydroxyl moiety being unfavorably located in a hydro-

phobic pocket. Thus, the fine specificity of the alloreactivity was

partly a consequence of the differential ability of HLA-B*4405,

HLA-B*4402, and HLA-B*4403 to accommodate plasticity of the

mimotope and allotope upon TCR ligation, further highlighting

the role and sensitivity of peptide-dependent molecular mimicry.

DISCUSSION

Molecular mimicry, namely when similar structures from dissim-

ilar proteins function in similar ways, is considered to underpin

receptor-ligand cross-reactivity in many biological systems

(Mariuzza and Poljak, 1993; Oldstone, 1987) and represents

a central tenet for therapeutic development of analog drugs.

Described in the 1980s in an immunological context (Williams,

I

1983), molecular mimicry is thought to be the basis for a number

of B cell autoimmune disorders, whereby the epitope from the

pathogen mimics the conformation of the self-ligand (Oldstone,

1987; Rose and Mackay, 2000). Evidence for molecular mimicry

of T cell ligands, though long suspected, has been harder to

establish structurally (Quaratino et al., 1995) because of the

dual specificity of T cell recognition for MHC and peptide. None-

theless, evidence is accumulating for mimicry as a basis of some

T cell autoimmunity (Harkiolaki et al., 2009; Hausmann et al.,

1999; Wucherpfennig and Strominger, 1995) and that T cell

cross-reactivity may be dependent on a few conserved germ-

line-encoded interactions (Dai et al., 2008). Here we describe

how extensive molecular mimicry underpins direct, human

T cell alloreactivity, a structurally unresolved phenomenon that

leads to tissue destruction and transplant rejection. In antiviral

immunity, small differences in the peptide or HLA molecule can

mmunity 31, 897–908, December 18, 2009 ª2009 Elsevier Inc. 905

Immunity

T Cell Alloreactivity Mediated by Molecular Mimicry

effectively ablate TCR corecognition of the viral determinant

bound to the HLA molecule. Thus, a priori it was unexpected

that a human T cell alloreaction between disparate HLA allotypes

was attributable to molecular mimicry. Indeed, a previously

described example of murine T cell alloreactivity showed how

the TCR adopted markedly different docking strategies when

recognizing self versus foreign ligands (Colf et al., 2007). To

exemplify this point further, HLA-B*4405 differs from HLA-

B*0801 by 25 amino acids in the Ag-binding cleft, of which 5 resi-

dues (positions 80, 82, 83, 163, and 167) are surface exposed and

potentially available for TCR contact. In the 2C TCR system of

alloreactivity, H2-Kb and H2-Ld molecules differ by 31 residues,

of which only 4 polymorphic residues are solvent exposed for

potentialTCRcontact (Colfetal.,2007).Moreover, ifoneconsiders

the apparent relatedness between the surface topologies of HLA-

B8 and HLA-B44, one would have anticipated that the LC13 TCR

alloreacts against all HLA-B44 allotopes, and this is clearly not

the case for HLA-B*4403, which differs from HLA-B*4405 by

only two buried polymorphic residues (Zernich et al., 2004).

In our study, the viral and allotrope peptides adopted similar

conformations only after binding the TCR. This induced-fit mech-

anism of molecular mimicry further explained why the TCR could

effectively discriminate between subtle polymorphic differences

between the foreign HLA-B*4402, HLA-B*4405, and HLA-

B*4403 allotypes. Thus, our data not only highlight the intricate

peptide dependence of T cell alloreactivity but also show that

direct T cell alloreactivity is attributable to exquisite specificity

of the TCR rather than degenerate recognition of MHC. Our find-

ings suggest that in transplantation, nonpermissive taboo

mismatches (Doxiadis et al., 1996) might depend on serendipi-

tous mimicry that is lacking in permissive mismatches.

Our observations in the LC13 TCR system and the contrasting

observations in the 2C TCR system raise the intriguing question

of whether molecular mimicry, or alternatively, disparate docking

modes between the cognate and allo-ligand will best explain the

general phenomenon of alloreactivity. The LC13 TCR system

describes alloreactivity between two disparate allotypes.

Because the LC13 TCR can alloreact via mimicry between these

two disparate allotypes, then it follows that alloreactivity

between more related alleles is likely to arise from mimicry.

Moreover, it also anticipated that molecular mimicry operates

between the alloreactions between HLA-B8 and HLA-B*3508

(Archbold et al., 2006) and between HLA-B*3508 and HLA-B44

(Tynan et al., 2005). Moreover, as T cells undergo thymic selec-

tion against self-pMHC, they are inherently cross-reactive, and

germline-encoded interactions are considered to underpin

MHC restriction (Scott-Browne et al., 2009); this further suggests

that mimicry will underpin most alloreactions. Consistent with

this, recent data (Dai et al., 2008; Rubtsova et al., 2009) suggest

that there are conserved CDR1/CDR2 interactions between the

cognate pMHC ligands and allogeneic-MHC class II molecules,

thereby indicating that molecular mimicry underpins this allore-

action. Nevertheless, more definitive data will be required

regarding the relative roles of mimicry versus disparate docking

modes typifying alloreactivity, and we suspect that there will be

a ‘‘sliding scale’’ between the two examples as suggested by the

2C TCR and LC13 TCR system.

To avoid auto-reactivity, the alloreactive LC13 clonotype

is absent from the peripheral T cell repertoire of HLA-

906 Immunity 31, 897–908, December 18, 2009 ª2009 Elsevier Inc.

B*0801+B*4402+ individuals (Burrows et al., 1995, 1997), instead

using an alternative T cell repertoire to recognize the HLA-

B8-restricted, FLR-virotope. Similar reshaping of the T cell

repertoire occurs in HLA-B*0801+B*4403+ heterozygotes

(Burrows et al., 1997), indicating that the LC13 clonotype is

sensitive to HLA-B*4403 in the thymus but not in the periphery.

This reflects the 30- to 100-fold increased ligand responsive-

ness of developing T cells compared with mature peripheral

T cells (Yagi and Janeway, 1990). Thus, although HLA-B*4403

appears to bind the allotope peptide less efficiently than

HLA-B*4405, this determinant is apparently still naturally pre-

sented with physiological consequences for the developing

T cell repertoire.

In line with molecular mimicry defining the LC13 alloreaction,

a prototypical TCR from the virotope-specific T cell repertoire

of HLA-B*0801+B*4402+ heterozygous individuals focused on

HLA-B*0801 residues that were polymorphic relative to HLA-

B44 (Gras et al., 2009). Molecular mimicry and aberrant T cell

reactivity represent important and long-standing themes in

immunology, and here these concepts converge to provide

a basis for understanding peptide-centric T cell alloreactivity.

EXPERIMENTAL PROCEDURES

Cell Lines and Transfectants

The class I reduced human B lymphoblastoid cell line Hmy2.C1R (C1R)

expresses HLA-Cw4 and has very low levels of HLA-B*3503 (Zemmour

et al., 1992). Transfection of C1R with HLA-B*4402, HLA-B*4403 (Macdonald

et al., 2003), or HLA-B*4405 with or without the herpes simplex virus TAP-

inhibitor ICP47 has been described previously (Zernich et al., 2004). BM.wil

is a human HLA-B*0801 homozygous, lymphoblastoid cell line transformed

with Epstein-Barr virus (EBV) (Charron, 1997) and constitutively expressing

low levels of EBV antigens including the FLRGRAYGL determinant (Burrows

et al., 1994). The T 3 B hybrid 174 3 CEM.T2 (T2) lacks TAP genes and its

transfectant derivatives, T2.B*0801 and T2.B*4405, express low levels of

‘‘empty’’ MHC-I gene products at the cell surface (Alexander et al., 1989;

Man et al., 1992). Jurkat.LC13 cells were generated by the retroviral transduc-

tion of the CD8a and b chains into Jurkat cells, as well as the LC13 TCR a and

b genes. The LC13 (Burrows et al., 1994) and DM1 (Archbold et al., 2009) anti-

viral CTL cones have been previously described.

Identification of Endogenous Alloligand

A randomized peptide library was engineered in complex with HLA-B*4405

molecules in a baculovirus vector (Crawford et al., 2006). The potential nona-

meric ‘‘allopeptide’’ library was constructed with random oligonucleotides but

fixing codons encoding P1E, a residue not likely to be involved in LC13 recog-

nition and the HLA-B*4405 anchor sites P2E and P9F or Y. PCR fragments

encoding the library were ligated to constructs encoding b2microglobulin

and the HLA-B*4405 heavy chain directing expression of individual HLA-

B*4405-peptide complexes from each virus. SF9 cells were infected with the

amplified viral stocks containing the HLA-B*4405-peptide library so that

each infected cell displayed a unique peptide-HLA complex. Cells were cos-

tained with fluoresceinated anti-b2Microglobulin, and fluorochrome-labeled

LC13 TCR ectodomain made multimeric with an anti-TCR mAb. Rare cells ex-

pressing HLA-B*4405-peptide complexes that bound LC13 TCR were repeat-

edly sorted and expanded by culture in vitro. After the 4th sorting, SF9 cells

homogeneously expressed a ligand that bound both anti-b2Microglobulin

and multimeric LC13.

T Cell Activation Assays

CTL killing assays (Burrows et al., 1994) and activation of Jurkat.LC13 cells

were assayed as previously described (Beddoe et al., 2009). Essentially,

Jurkat.LC13 cells (105) were cocultured with 105 antigen-presenting cells for

4 hr at 37�C in the absence or presence of peptide. Expression of CD69 was

Immunity

T Cell Alloreactivity Mediated by Molecular Mimicry

then detected by flow cytometry gating on GFP-positive LC13.Jurkat cells.

T cell activation was measured as the percentage of CD69-positive cells

among the GFP-positive LC13.Jurkat cells relative to the unstimulated popu-

lation. RNAi knockdown of ABCD3 is described in the Supplemental Data.

Primary T cells were obtained from blood donors with the approval of the

Australian Bone Marrow Donor Registry Ethics Committee Scientific Review

Panel.

Additional Data

Supplemental Data include protein expression, purification, crystallization,

structure determination, and SPR measurement.

ACCESSION NUMBERS

Coordinates have been deposited in the PDB (codes: 3KPL, 3KPM, 3KPN,

3KPO, 3KPP, 3KPQ, 3KPS, 3KPR).

SUPPLEMENTAL DATA

Supplemental Data include Supplemental Experimental Procedures, seven

figures, and three tables and can be found with this article online at http://

www.cell.com/immunity/supplemental/S1074-7613(09)00510-X.

ACKNOWLEDGMENTS

We thank the staff at the GMCA-CAT beamline (Chicago, Advanced Photon

Source) and the PX1 Beamline at the Australian synchrotron for assistance

with data collection. We thank Hugh Reid and Kim R. Jordan for technical

advice. J.R. is a Federation Fellow of the Australian Research Council,

C.S.C. is an ARC QEII fellow, W.A.M. is a Peter Doherty Postdoctoral Fellow,

F.E.T. is a CJ Martin Fellow, and A.W.P., M.C.J.W., and S.R.B. are Senior

Research Fellows of the National Health and Medical Research Council

Australia. This work was supported by grants from the ARC, NHMRC, and

Roche Organ Transplant Research Foundation.

Received: June 10, 2009

Revised: September 10, 2009

Accepted: September 25, 2009

Published online: December 17, 2009

REFERENCES

Afzali, B., Lechler, R.I., and Hernandez-Fuentes, M.P. (2007). Allorecognition

and the alloresponse: clinical implications. Tissue Antigens 69, 545–556.

Alexander, J., Payne, J.A., Murray, R., Frelinger, J.A., and Cresswell, P. (1989).

Differential transport requirements of HLA and H-2 class I glycoproteins.

Immunogenetics 29, 380–388.

Archbold, J.K., Macdonald, W.A., Miles, J.J., Brennan, R.M., Kjer-Nielsen, L.,

McCluskey, J., Burrows, S.R., and Rossjohn, J. (2006). Alloreactivity

between disparate cognate and allogeneic pMHC-I complexes is the result

of highly focused, peptide-dependent structural mimicry. J. Biol. Chem. 281,

34324–34332.

Archbold, J.K., Ely, L.K., Kjer-Nielsen, L., Burrows, S.R., Rossjohn, J., McClus-

key, J., and Macdonald, W.A. (2008a). T cell allorecognition and MHC restric-

tion—A case of Jekyll and Hyde? Mol. Immunol. 45, 583–598.

Archbold, J.K., Macdonald, W.A., Burrows, S.R., Rossjohn, J., and McClus-

key, J. (2008b). T-cell allorecognition: A case of mistaken identity or deja

vu? Trends Immunol. 29, 220–226.

Archbold, J.K., Macdonald, W.A., Gras, S., Ely, L.K., Miles, J.J., Bell, M.J.,

Brennan, R.M., Beddoe, T., Wilce, M.C.J., Clements, C.S., et al. (2009). Natural

micropolymorphism in human leukocyte antigens provides a basis for genetic

control of antigen recognition. J. Exp. Med. 206, 209–219.

Argaet, V.P., Schmidt, C.W., Burrows, S.R., Silins, S.L., Kurilla, M.G., Doolan,

D.L., Suhrbier, A., Moss, D.J., Kieff, E., Sculley, T.B., and Misko, I.S. (1994).

Dominant selection of an invariant T cell antigen receptor in response to persis-

tent infection by Epstein-Barr virus. J. Exp. Med. 180, 2335–2340.

I

Beddoe, T., Chen, Z., Clements, C.S., Ely, L.K., Bushell, S.R., Vivian, J.P., Kjer-

Nielsen, L., Pang, S.S., Dunstone, M.A., Liu, Y.C., et al. (2009). Antigen ligation

triggers a conformational change within the constant domain of the alphabeta

T cell receptor. Immunity 30, 777–788.

Bevan, M.J. (1984). High determinant density may explain the phenomenon of

alloreactivity. Immunol. Today 5, 128–130.

Bevan, M.J., and Hunig, T. (1981). T cells respond preferentially to antigens

that are similar to self H-2. Proc. Natl. Acad. Sci. USA 78, 1843–1847.

Borg, N.A., Ely, L.K., Beddoe, T., Macdonald, W.A., Reid, H.H., Clements,

C.S., Purcell, A.W., Kjer-Nielsen, L., Miles, J.J., Burrows, S.R., et al. (2005).

The CDR3 regions of an immunodominant T cell receptor dictate the ‘energetic

landscape’ of peptide-MHC recognition. Nat. Immunol. 6, 171–180.

Borg, N.A., Wun, K.S., Kjer-Nielsen, L., Wilce, M.C., Pellicci, D.G., Koh, R.,

Besra, G.S., Bharadwaj, M., Godfrey, D.I., McCluskey, J., and Rossjohn, J.

(2007). CD1d-lipid-antigen recognition by the semi-invariant NKT T-cell

receptor. Nature 448, 44–49.

Burrows, S.R., Khanna, R., Burrows, J.M., and Moss, D.J. (1994). An allores-

ponse in humans is dominated by cytotoxic T lymphocytes (CTL) cross-

reactive with a single Epstein-Barr virus CTL epitope: implications for graft-

versus-host disease. J. Exp. Med. 179, 1155–1161.

Burrows, S.R., Silins, S.L., Moss, D.J., Khanna, R., Misko, I.S., and Argaet, V.P.

(1995). T cell receptor repertoire for a viral epitope in humans is diversified by

tolerance to a background major histocompatibility complex antigen. J. Exp.

Med. 182, 1703–1715.

Burrows, S.R., Silins, S.L., Cross, S.M., Peh, C.A., Rischmueller, M., Burrows,

J.M., Elliott, S.L., and McCluskey, J. (1997). Human leukocyte antigen pheno-

type imposes complex constraints on the antigen-specific cytotoxic T lympho-

cyte repertoire. Eur. J. Immunol. 27, 178–182.

Charron, D. (1997). Genetic Diversity of HLA Functional and Medical Implica-

tion, Volume 1 (Paris: EDK Medical and Scientific International Publisher).

Colf, L.A., Bankovich, A.J., Hanick, N.A., Bowerman, N.A., Jones, L.L., Kranz,

D.M., and Garcia, K.C. (2007). How a single T cell receptor recognizes both self

and foreign MHC. Cell 129, 135–146.

Crawford, F., Jordan, K.R., Stadinski, B., Wang, Y., Huseby, E., Marrack, P.,

Slansky, J.E., and Kappler, J.W. (2006). Use of baculovirus MHC/peptide

display libraries to characterize T-cell receptor ligands. Immunol. Rev. 210,

156–170.

Dai, S., Huseby, E.S., Rubtsova, K., Scott-Browne, J., Crawford, F., Macdon-

ald, W.A., Marrack, P., and Kappler, J.W. (2008). Crossreactive T Cells spot-

light the germline rules for alphabeta T cell-receptor interactions with MHC

molecules. Immunity 28, 324–334.

Davis, M.M., Boniface, J.J., Reich, Z., Lyons, D., Hampl, J., Arden, B., and

Chien, Y. (1998). Ligand recognition by alpha beta T cell receptors. Annu.

Rev. Immunol. 16, 523–544.

Doxiadis, I.I., Smits, J.M., Schreuder, G.M., Persijn, G.G., van Houwelingen,

H.C., van Rood, J.J., and Claas, F.H. (1996). Association between specific

HLA combinations and probability of kidney allograft loss: The taboo concept.

Lancet 348, 850–853.

Droge, W. (1979). Hypothesis on the origin of the strong alloreactivity. Immu-

nobiology 156, 2–12.

Felix, N.J., Donermeyer, D.L., Horvath, S., Walters, J.J., Gross, M.L., Suri, A.,

and Allen, P.M. (2007). Alloreactive T cells respond specifically to multiple

distinct peptide-MHC complexes. Nat. Immunol. 8, 388–397.

Fink, P.J., and Bevan, M.J. (1995). Positive selection of thymocytes. Adv.

Immunol. 59, 99–133.

Fleischhauer, K., Kernan, N.A., O’Reilly, R.J., Dupont, B., and Yang, S.Y.

(1990). Bone marrow-allograft rejection by T lymphocytes recognizing a single

amino acid difference in HLA-B44. N. Engl. J. Med. 323, 1818–1822.

Garboczi, D.N., and Biddison, W.E. (1999). Shapes of MHC restriction. Immu-

nity 10, 1–7.

Garcia, K.C., Degano, M., Pease, L.R., Huang, M., Peterson, P.A., Teyton, L.,

and Wilson, I.A. (1998). Structural basis of plasticity in T cell receptor recogni-

tion of a self peptide-MHC antigen. Science 279, 1166–1172.

mmunity 31, 897–908, December 18, 2009 ª2009 Elsevier Inc. 907

Immunity

T Cell Alloreactivity Mediated by Molecular Mimicry

Garcia, K.C., Degano, M., Speir, J.A., and Wilson, I.A. (1999). Emerging

principles for T cell receptor recognition of antigen in cellular immunity. Rev.

Immunogenet. 1, 75–90.

Gras, S., Burrows, S.R., Kjer-Nielsen, L., Clements, C.S., Liu, Y.C., Sullivan,

L.C., Bell, M.J., Brooks, A.G., Purcell, A.W., McCluskey, J., and Rossjohn, J.

(2009). The shaping of T cell receptor recognition by self-tolerance. Immunity

30, 193–203.

Harkiolaki, M., Holmes, S.L., Svendsen, P., Gregersen, J.W., Jensen, L.T.,

McMahon, R., Friese, M.A., van Boxel, G., Etzensperger, R., Tzartos, J.S.,

et al. (2009). T cell-mediated autoimmune disease due to low-affinity cross-

reactivity to common microbial peptides. Immunity 30, 348–357.

Hausmann, S., Martin, M., Gauthier, L., and Wucherpfennig, K.W. (1999).

Structural features of autoreactive TCR that determine the degree of degen-

eracy in peptide recognition. J. Immunol. 162, 338–344.

Heath, W.R., Hurd, M.E., Carbone, F.R., and Sherman, L.A. (1989). Peptide-

dependent recognition of H-2Kb by alloreactive cytotoxic T lymphocytes.

Nature 341, 749–752.

Heath, W.R., Kane, K.P., Mescher, M.F., and Sherman, L.A. (1991). Alloreac-

tive T cells discriminate among a diverse set of endogenous peptides. Proc.

Natl. Acad. Sci. USA 88, 5101–5105.

Heemskerk, M.B., Cornelissen, J.J., Roelen, D.L., van Rood, J.J., Claas, F.H.,

Doxiadis, I.I., and Oudshoorn, M. (2007). Highly diverged MHC class I

mismatches are acceptable for haematopoietic stem cell transplantation.

Bone Marrow Transplant. 40, 193–200.

Hogquist, K.A., Gavin, M.A., and Bevan, M.J. (1993). Positive selection of

CD8+ T cells induced by major histocompatibility complex binding peptides

in fetal thymic organ culture. J. Exp. Med. 177, 1469–1473.

Jameson, S.C., Hogquist, K.A., and Bevan, M.J. (1995). Positive selection of

thymocytes. Annu. Rev. Immunol. 13, 93–126.

Kawase, T., Morishima, Y., Matsuo, K., Kashiwase, K., Inoko, H., Saji, H., Kato,

S., Juji, T., Kodera, Y., Sasazuki, T., and Japan Marrow Donor Program. (2007).

High-risk HLA allele mismatch combinations responsible for severe acute

graft-versus-host disease and implication for its molecular mechanism. Blood

110, 2235–2241.

Keever, C.A., Leong, N., Cunningham, I., Copelan, E.A., Avalos, B.R., Klein, J.,

Kapoor, N., Adams, P.W., Orosz, C.G., Tutschka, P.J., et al. (1994). HLA-B44-

directed cytotoxic T cells associated with acute graft-versus-host disease

following unrelated bone marrow transplantation. Bone Marrow Transplant.

14, 137–145.

Kjer-Nielsen, L., Clements, C.S., Brooks, A.G., Purcell, A.W., Fontes, M.R.,

McCluskey, J., and Rossjohn, J. (2002a). The structure of HLA-B8 complexed

to an immunodominant viral determinant: peptide-induced conformational

changes and a mode of MHC class I dimerization. J. Immunol. 169, 5153–5160.

Kjer-Nielsen, L., Clements, C.S., Brooks, A.G., Purcell, A.W., McCluskey, J.,

and Rossjohn, J. (2002b). The 1.5 A crystal structure of a highly selected anti-

viral T cell receptor provides evidence for a structural basis of immunodomi-

nance. Structure 10, 1521–1532.

Kjer-Nielsen, L., Clements, C.S., Purcell, A.W., Brooks, A.G., Whisstock, J.C.,

Burrows, S.R., McCluskey, J., and Rossjohn, J. (2003). A structural basis for

the selection of dominant alphabeta T cell receptors in antiviral immunity.

Immunity 18, 53–64.

Lechler, R., and Lombardi, G. (1990). The structural basis of alloreactivity.

Immunol. Res. 9, 135–146.

Lindahl, K.F., and Wilson, D.B. (1977). Histocompatibility antigen-activated

cytotoxic T lymphocytes. I. Estimates of the absolute frequency of killer cells

generated in vitro. J. Exp. Med. 145, 500–507.

Macdonald, W.A., Purcell, A.W., Mifsud, N.A., Ely, L.K., Williams, D.S., Chang,

L., Gorman, J.J., Clements, C.S., Kjer-Nielsen, L., Koelle, D.M., et al. (2003).

A naturally selected dimorphism within the HLA-B44 supertype alters class I

structure, peptide repertoire, and T cell recognition. J. Exp. Med. 198, 679–691.

Man, S., Salter, R.D., and Engelhard, V.H. (1992). Role of endogenous peptide

in human alloreactive cytotoxic T cell responses. Int. Immunol. 4, 367–375.

Mariuzza, R.A., and Poljak, R.J. (1993). The basics of binding: Mechanisms of

antigen recognition and mimicry by antibodies. Curr. Opin. Immunol. 5, 50–55.

908 Immunity 31, 897–908, December 18, 2009 ª2009 Elsevier Inc.

Matzinger, P., and Bevan, M.J. (1977). Hypothesis: Why do so many lympho-

cytes respond to major histocompatibility antigens? Cell. Immunol. 29, 1–5.

Mifsud, N.A., Purcell, A.W., Chen, W., Holdsworth, R., Tait, B.D., and McClus-

key, J. (2008). Immunodominance hierarchies and gender bias in direct

T(CD8)-cell alloreactivity. Am. J. Transplant. 8, 121–132.

Oldstone, M.B. (1987). Molecular mimicry and autoimmune disease. Cell 50,

819–820.

Parham, P., and Ohta, T. (1996). Population biology of antigen presentation by

MHC class I molecules. Science 272, 67–74.

Quaratino, S., Thorpe, C.J., Travers, P.J., and Londei, M. (1995). Similar

antigenic surfaces, rather than sequence homology, dictate T-cell epitope

molecular mimicry. Proc. Natl. Acad. Sci. USA 92, 10398–10402.

Reiser, J.B., Darnault, C., Guimezanes, A., Gregoire, C., Mosser, T., Schmitt-

Verhulst, A.M., Fontecilla-Camps, J.C., Malissen, B., Housset, D., and Mazza,

G. (2000). Crystal structure of a T cell receptor bound to an allogeneic MHC

molecule. Nat. Immunol. 1, 291–297.

Rose, N.R., and Mackay, I.R. (2000). Molecular mimicry: a critical look at

exemplary instances in human diseases. Cell. Mol. Life Sci. 57, 542–551.

Rossjohn, J., and McCluskey, J. (2007). How a home-grown T cell receptor

interacts with a foreign landscape. Cell 129, 19–20.

Rubtsova, K., Scott-Browne, J.P., Crawford, F., Dai, S., Marrack, P., and Kap-

pler, J.W. (2009). Many different Vbeta CDR3s can reveal the inherent MHC

reactivity of germline-encoded TCR V regions. Proc. Natl. Acad. Sci. USA

106, 7951–7956.

Rudolph, M.G., Stanfield, R.L., and Wilson, I.A. (2006). How TCRs bind MHCs,

peptides, and coreceptors. Annu. Rev. Immunol. 24, 419–466.

Scott-Browne, J.P., White, J., Kappler, J.W., Gapin, L., and Marrack, P. (2009).

Germline-encoded amino acids in the alphabeta T-cell receptor control thymic

selection. Nature 458, 1043–1046.

Sherman, L.A., and Chattopadhyay, S. (1993). The molecular basis of allore-

cognition. Annu. Rev. Immunol. 11, 385–402.

Smith, K.D., Huczko, E., Engelhard, V.H., Li, Y.Y., and Lutz, C.T. (1997a). Allor-

eactive cytotoxic T lymphocytes focus on specific major histocompatibility

complex-bound peptides. Transplantation 64, 351–359.

Smith, P.A., Brunmark, A., Jackson, M.R., and Potter, T.A. (1997b). Peptide-

independent recognition by alloreactive cytotoxic T lymphocytes (CTL). J.

Exp. Med. 185, 1023–1033.

Speir, J.A., Garcia, K.C., Brunmark, A., Degano, M., Peterson, P.A., Teyton, L.,

and Wilson, I.A. (1998). Structural basis of 2C TCR allorecognition of H-2Ld

peptide complexes. Immunity 8, 553–562.

Tynan, F.E., Burrows, S.R., Buckle, A.M., Clements, C.S., Borg, N.A., Miles,

J.J., Beddoe, T., Whisstock, J.C., Wilce, M.C., Silins, S.L., et al. (2005).

T cell receptor recognition of a ‘super-bulged’ major histocompatibility

complex class I-bound peptide. Nat. Immunol. 6, 1114–1122.

Williams, R.C., Jr. (1983). Rheumatic fever and the streptococcus. Another

look at molecular mimicry. Am. J. Med. 75, 727–730.

Wucherpfennig, K.W., and Strominger, J.L. (1995). Molecular mimicry in T cell-

mediated autoimmunity: viral peptides activate human T cell clones specific

for myelin basic protein. Cell 80, 695–705.

Yagi, J., and Janeway, C.A., Jr. (1990). Ligand thresholds at different stages of

T cell development. Int. Immunol. 2, 83–89.

Zemmour, J., Little, A.M., Schendel, D.J., and Parham, P. (1992). The HLA-A,B

‘‘negative’’ mutant cell line C1R expresses a novel HLA-B35 allele, which

also has a point mutation in the translation initiation codon. J. Immunol. 148,

1941–1948.

Zernich, D., Purcell, A.W., Macdonald, W.A., Kjer-Nielsen, L., Ely, L.K., Laham,

N., Crockford, T., Mifsud, N.A., Bharadwaj, M., Chang, L., et al. (2004). Natural

HLA class I polymorphism controls the pathway of antigen presentation and

susceptibility to viral evasion. J. Exp. Med. 200, 13–24.

Zinkernagel, R.M., and Doherty, P.C. (1974). Restriction of in vitro T cell-medi-

ated cytotoxicity in lymphocytic choriomeningitis within a syngeneic or semi-

allogeneic system. Nature 248, 701–702.